Wan 2.2

All the models available for video generation.

ImagineArt Model Help Guide: Wan 2.2

---

Introduction

Wan 2.2 is the latest evolution in the Wan AI series, bringing advanced capabilities for both text-to-video (T2V) and image-to-video (I2V) creation. With its new Mixture-of-Experts (MoE) diffusion architecture, larger training dataset, and 5B hybrid model, Wan 2.2 provides smoother motion, richer aesthetics, and more precise prompt adherence than its predecessor, Wan 2.1. Whether you're creating cinematic videos or artistic visuals, Wan 2.2 offers enhanced flexibility and control, enabling you to bring your creative ideas to life more effectively. This guide will walk you through how to use Wan 2.2, how to craft high-quality prompts, and how it compares to other models on ImagineArt.

---

Key Upgrades from Wan 2.1

Feature	Wan 2.1	Wan 2.2
Core Architecture	Dense diffusion	MoE diffusion with expert hand-off
Training Data	Baseline set	+65.6% images, +83.2% videos
Aesthetic Control	Basic tags	Cinematic-level controls
Motion Fidelity	Moderate	Complex, stable multi-object motion
Model Line-up	14B T2V, I2V	14B T2V/I2V + 5B hybrid TI2V

Why This Matters:

• Sharper Frames & Fewer Artifacts: The MoE diffusion and larger dataset improve the video’s overall sharpness and clarity.

• Smoother, More Controllable Motion: The ability to handle complex and stable multi-object motion creates more fluid and dynamic scenes.

• Hybrid Model for Local Prototyping: The 5B hybrid model enables powerful video generation locally without requiring heavy computational resources.

---

How to Access Wan 2.2 in ImagineArt

To use Wan 2.2, follow these simple steps:

1. Log into ImagineArt.

2. Open the AI Video Generator.

3. Select Wan 2.2 from the available model list.

4. You can choose from Wan 2.2 Lite or Wan 2.2 Pro based on your project’s needs.

---

How to Use Wan 2.2 in ImagineArt: Step-by-Step Guide

1. Open the AI Video Generator.

2. Select Wan 2.2 from the available models.

3. Upload reference images or videos if desired (note that only the starting frame of videos will be used).

4. Write your prompt specifying the details of the video you want to generate.

5. Set your desired aspect ratio, resolution, and duration (typically 5–10 seconds).

6. Open Advanced Settings to tweak seeds for more precise control over the final output.

7. Click Generate to start the rendering process.

Within a few seconds, Wan 2.2 will process your prompt and generate the video, ready for download or further iteration.

---

Prompt Writing Formulas

Effective prompts are crucial for achieving high-quality video outputs with Wan 2.2. Here’s how to craft the best prompts using both basic and advanced structures:

---

Basic Formula

The Basic Formula is ideal for quick, straightforward prompts. It focuses on the core components of the video:

• Subject: The main object (e.g., a person, animal, or object).

• Scene: The environment or background details.

• Motion: The movement of the subject or objects.

Example:

• "A red fox trotting slowly through a snowy pine forest as snowflakes fall around it."

---

Advanced Formula

The Advanced Formula adds more details and offers greater control over aesthetics and scene composition:

• Subject Description: Specific appearance of the subject.

• Scene Description: Adjectives describing the environment.

• Motion Description: Details about movement speed, amplitude, and effects.

• Aesthetic Control: Lighting, camera type, lens, time of day.

• Stylization: Artistic style (e.g., cyberpunk, watercolor, etc.).

Example:

• "A black-haired Miao girl in traditional dress, standing by a river under lanterns. Camera dolly-in, warm rim light, Kodak Portra grade, cinematic bokeh."

---

Image-to-Video Formula

For image-to-video generation, use a high-quality visual reference, and then describe the motion and camera movement:

• Motion: What changes or moves in the scene.

• Camera Movement: e.g., push-in, pan left, fixed camera.

Example:

• "The knight raises his sword slowly, camera pulls back to reveal a vast battlefield."

---

Advanced Prompt Techniques

Take your videos to the next level with these advanced prompting techniques:

• Shot Order: Start with the opening shot, then describe camera motion, and conclude with a pay-off shot.

• Camera Language: Use terms like pan, tilt, dolly, orbit, crane to describe camera movement.

• Motion Modifiers: Add specific modifiers like slow-motion, whip-pan, or fixed elements in the background.

• Aesthetic Tags: Enhance your videos by adding details like lighting (e.g., volumetric dusk, neon rim light), color schemes (e.g., teal-and-orange), or lens effects (e.g., anamorphic bokeh).

• Negative Prompting: Avoid undesired elements such as "bright colors," "overexposed," "low-quality," "extra fingers," or "malformed limbs."

---

Sample Wan 2.2 Prompts

Here are a few examples to inspire your prompts:

1. Neon Drift (Cyberpunk Tracking Shot)

   Prompt: "A rainy night in a dense cyberpunk market, neon signs flicker overhead. Camera tracks behind a hooded courier weaving through crowds. Volumetric pink-blue backlight cuts through steam vents, puddles reflect the glow."

2. Alpine Reveal (Pull Back)

   Prompt: "Extreme close-up of a mountaineer’s ice axe biting into frozen rock. Camera pulls back to reveal the climber and a vast sunrise-lit alpine ridge."

3. Aquatic Ballet (Slow Motion Orbit)

   Prompt: "An orca breaches in crystal-clear Arctic waters. Slow 360° orbital shot around the whale, droplets hang suspended. Pastel polar sunset lighting."

---

Wan 2.2 vs. Other AI Video Models

Here’s a comparison of Wan 2.2 with other AI video models available on ImagineArt:

Feature	Wan 2.2	Google Veo 3	Runway Gen-4	MiniMax Hailuo 02	Kling 2.1
Visual Quality	720p @ 24fps	720p	720p	512p/768p/1080p	720p/1080p
Video Length	5–10 seconds	8 seconds	5–10 seconds	6 seconds	5–10 seconds
Audio Generation	No	Yes	No	No	No
Lip Sync	Not supported	Native	Not supported	Not supported	Not supported
Prompt Input	Text + Image/Video	Text + Image	Text + Image	Text + Image	Text + Image
Multi-shot Consistency	Complex, stable	Limited	Limited	Basic	Limited
Camera Control	Cinematic styles	Prompt-based	Stylized transitions	Manual presets	Basic

---

Conclusion: Is Wan 2.2 Worth Trying?

Wan 2.2 is an advanced AI video model that offers sharp, stable video outputs, superior motion fidelity, and cinematic-level aesthetic controls. The MoE diffusion architecture and 5B hybrid model make it ideal for both large-scale productions and lightweight local prototyping. If you need enhanced control over video aesthetics and multi-object motion, Wan 2.2 is a fantastic choice. Try it today in the ImagineArt AI Video Generator and create visually stunning content effortlessly!